>
> With SQL Server, all it does is tell it to pass the data back before the
> previous locking status - so, instead of checking each row to see if it's
> locked, it passes the "current" data - if you're searching a very large
> table, then what does it matter if you're 20-30 seconds out of date
> I've found that the speed increase of using NOLOCK is around 30% - if
that's
> not worth your while, then you must be running one hell of a fast database
> engine <g>
> Do some tests returning several thousand records and check the query
speed -
> I can (virtually) guarantee that with NOLOCK you'll get a marked
improvement
> for the loss of the very latest updates...
>
That's interesting. I didn't know that NOLOCK could increase performance.
I see your point about the trade off too. I guess it would depend on the
application. My current project is purely reporting on data that is being
inserted, but never updated....so using NOLOCK should be benign. However,
my previous work was on a very large application with potentially hundreds
of different processes inserting, updating, and inserting data in "real
time". In this scenario, the use of NOLOCK carries an unacceptable amount
of risk with it.
Brian
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Structure your ColdFusion code with Fusebox. Get the official book at
http://www.fusionauthority.com/bkinfo.cfm
FAQ: http://www.thenetprofits.co.uk/coldfusion/faq
Archives: http://www.mail-archive.com/cf-talk@houseoffusion.com/
Unsubscribe: http://www.houseoffusion.com/index.cfm?sidebar=lists
